Sabitlenmiş Tweet
Souei
689 posts

Souei
@SS_Soeui
Mehn I'm just soo tired maybe Sol would keep me going
Nowhere near Katılım Nisan 2026
33 Takip Edilen23 Takipçiler

FTX’s former law firm and auditor agree to $66M settlement
Fenwick & West will pay $54M and auditor Prager Metis will pay $11.75M as part of a second wave of FTX customer settlements tied to fraud claims involving Sam Bankman-Fried.
The settlement group also includes former NBA player Udonis Haslem, while Fenwick still faces a separate $525M lawsuit.

English

@CryptoMiners_Co At least some money’s coming back to users. Most of these cases drag for years and end with nothing
English

this challenge was chosen by my TG community 👀
if you want to be part of the fam, join here 👇🏻
i also do TG-exclusive giveaways
t.me/PurpleHoodieCo…
English

Mines 500x challenge is officially completed ✅
@amredox hit it and claimed the $100
@DiceyHQ Week is just getting started tho
next challenge TODAY 👀🎲

chrizzy@chrizzy
nobody hit the 1000x Mines challenge yet, so let’s make it more interesting lowering it to 500x first person under code CHRIZZY to hit it on @DiceyHQ Mines gets $100 min bet: $0.20 send proof when you hit GL 🎲
English

I’ve tried for this entire past year to use subagents inside agents like Codex / Claude Code. At the moment, I can tell you that yes, when the models are at peak performance, when they’re first released, the result is very good. But already after a week, when everything drops off as if it were a rug pull, the orchestration is no longer actually followed the way it was at the beginning.
Over this year, I’ve tried everything to optimize my workflow with these agents, trying to ship more and more products in the shortest possible time. But the reality is that, at the moment, it cannot be done. To avoid ending up with slop code, you need to use only one single session, without compaction, to perform updates. It will take you more time, but afterward you definitely won’t have to go back and fix bugs caused by context loss.
Because of context loss and memory storage issues, I’ve tried countless things. As I’m testing, and as I’ll continue to test on my recently released open-source project, you can push the model to the very limit of its orchestration abilities. But if after a while that much-touted 5–10% improvement from benchmarks causes everything to break, then nothing follows a workflow anymore that would allow subagents to correctly integrate something or properly fix an issue they’ve been given. This is because subagents seem to have far less context, much, much less, than the main agents that spawn them. They genuinely seem dumber, despite the fact that I built a system to pass all the necessary information both as input and output in order to carry out a specific update.
Subagents are not like a normal session. They are clearly debuffed compared to the main session. Most likely, once the model opens the subagent, it doesn’t work as it should, and therefore performs many actions that make clean orchestration impossible, for example, running the file-reading tool 10 times, while in the main session it reads the file with a single run.
English

Squid raises $6M with Ripple backing cross-chain expansion
@squidrouter has raised $6M in a strategic round led by North Island Ventures with participation from @Ripple, Dialectic, and Borderless, bringing total funding to $13.5M.
The protocol has processed over $6B across 4M transactions for more than 1M users since 2023, connecting 100+ chains and 20,000+ tokens while serving as an XRP Ledger bridge partner and validator.

English

@CryptoMiners_Co @squidrouter @Ripple Another bridge raise in 2025 though? Hope they’ve fixed the exploit risk better than most
English

I'm Back
To celebrate my return to CT, I'm doing a $1,000 giveaway
How to enter:
- Like & RT this post
- Tag 2 friends
- Follow @Trident100x with Notis on 🔔
- Drop your $SOL wallet in the comments
Winner picked in 72 hours 👀

English

EURR and USDR lose peg after StablR exploit mints $13.5M in unbacked tokens
@StablREuro’s EURR and USDR stablecoins depegged after an attacker compromised the issuer’s 1-of-3 minting multisig, minting 4.5M EURR and 8.35M USDR before dumping roughly $10.4M across decentralized exchanges.
EURR fell to around $0.85 while USDR dropped as low as $0.40 following the exploit.

English

@CryptoMiners_Co @StablREuro This is why ‘stable’ feels like a joke unless it’s fully audited and overcollateralized. Depeg risk is real
English

Vitalik says Ethereum Foundation will become a “smaller ship” and sell less ETH
Vitalik Buterin said the Ethereum Foundation is prioritizing longevity over breadth, reducing ETH sales, and focusing on censorship resistance, open source development, privacy, and security.
The comments come amid researcher departures, with Buterin also saying his influence within the foundation will continue to decrease over time.

English

@CryptoMiners_Co Censorship resistance and privacy focus hits different after the last few cycles. Glad they’re doubling down
English

Nigga Ray J very publicly exposed himself for rigging a fight for betting profits and was raging when his opponent didn't follow through with their end of the deal 😂😂
It's no wonder he was begging Sahil for som BTC on a dirty shitcoin shill.
sahil@villainmonkey
Ray J 😭
English

When you login to @acebet for the day with a fat balance.
Let's cook!
Use code: "Thedefiape" to compete for prizes on my wager leaderboard.
acebet.co/welcome/r/thed…
English


